With the Honda BENLY e: electric scooter series, Honda wants to take a significant step forward in urban mobility. These electric scooters are primarily designed for business use and are intended to meet the requirements of delivery services and other commercial applications.
As of November 2023, Honda has set itself an annual sales target of four million electric scooters for 2030 in order to take a further step towards climate neutrality. The power control unit (PCU) of the scooters is developed by Honda itself. For the final tests of the PCU at full power, Honda relies on dSPACE power hardware-in-the-loop (HIL) technology to ensure comprehensive validation, product safety, and the avoidance of bottlenecks in the development process.
Power Control Unit as the Centerpiece
"At the heart of the scooter is a regulated PCU, which is powered by two battery packs and drives a three-phase, permanent magnet synchronous motor," explains Etsuko Tokunaga from Honda. In addition to motor control, the PCU also has the task of maintaining the scooter's power limits, providing the 12 V auxiliary power supply, e.g., for sensors, and monitoring all components for any faults that occur.
Numerous test cases must be covered, particularly in the case of comprehensive validation, in order to ensure that the scooter functions reliably even in the event of a failure. This applies in particular to the PCU, which is validated with the dSPACE power HIL system. “This enables testing of under the conditions that closely simulate real-world usage, and at the same time reproduce and confirm variable test cases and environmental conditions in a matter of seconds,” says Tokunaga, explaining the advantages.
Simulation of the Battery Packs and the Motor at Full Power
The primary role of the dSPACE power HIL system is to simulate the behavior of the battery packs and the three-phase motor, enabling testing at power levels equivalent to those in real-world applications. This allows the PCU to be embedded in the electrical environment of the drivetrain without the need for the actual battery and motor components to be available. This prevents bottlenecks in the validation process, as the PCU can be tested immediately after assembly.
Because the same currents occur in the power HIL environment as in the real environment with a real motor and battery, Honda can also use the actual current sensors in the test. All other sensor and position encoder signals and CAN commands required by the PCU are provided by the power HIL system.
Overview diagram of the power HIL validation setup, in which real components are replaced by an emulated environment
Accurate Models Are Crucial
The power HIL system is based on three main components. First, the environmental behavior of the considered battery and the corresponding motor must be simulated as accurately as possible in real time. Honda relies on preconfigured battery and motor models for processor or FPGA. These can be used directly and parameterized as desired. Alternatively, the open model structure based on MATLAB Simulink also offers full freedom to change the model at a later date as desired or even to implement custom models, provided that these can still be executed in real time.
In the battery model, in addition to CAN communication between the battery and PCU, the focus is on precise emulation of cell voltages and cell temperatures. The motor model simulates both the electrical and mechanical variables of the motor, such as current, voltage, torque, and speed, as well as a position encoder/resolver signal. In addition, both models implement configurable fault cases to verify whether the PCU can safely control the scooter in the event of a failure.
The second component is a real-time system that executes the models in real time and transmits the corresponding signals and target values to the PCU and power HIL power electronics. A SCALEXIO real-time processor calculates the battery model. At the same time, it serves as an interface to the test bench's control computer to set operating points and parameters and read out measurement data. The much more complex machine model is implemented on an FPGA. This is housed in a SCALEXIO LabBox, in which all the necessary signal interfaces are also made available in the form of plug-in cards. A signal-based HIL test is already possible with this setup.
Full-Powered Load Modules
High-voltage load modules are the third component to raise the model behavior to the power level. Each module is equipped with a fast-switching, multi-level output stage with silicon carbide MOSFETs and decentralized current control. Each module can behave as a highly dynamic regenerative power supply, which only receives the setpoint values from the real-time system in order to convert model variables into electrical variables. Output voltages of up to 1,250 V and output currents of up to 75 ARMS are possible per module without power limitation.
While the voltage limit represents a fixed upper limit due to the Low Voltage Directive, the maximum current and thus the maximum power of the system can be scaled upwards as required by connecting several load modules in parallel. “In the BENLY e: test bench, each of the three machine phases is simulated with multiple load modules connected in parallel, while the positive and negative terminals of the battery have multiple load modules,” details Etsuko Tokunaga. The relatively low battery voltage of around 100 V also poses no challenge for the downward-compatible load modules.
Detect Software and Hardware Problems at an Early Stage Without Risk
With this setup, comprehensive and reproducible tests of the PCU will be possible as soon as the first prototype is available. Software and hardware problems can now be detected at an early stage without having to wait for further drivetrain components to be manufactured. The overall schedule is not delayed unnecessarily in the event of an error.
Etsuko Tokunaga emphasizes: "In particular, the simple parameterization of the battery and motor parameters makes it possible to cover a large number of test cases without having to take physical restrictions on the rate of change into account." In the case of the virtual battery, it can be charged and discharged in fractions of a second at the click of a mouse.
With this flexibility, fault validation is easy to implement. The framework conditions for an undervoltage test of the batteries can be parameterized in the same way as those for an overvoltage test with overcharging of the battery. While the latter poses a risk of destroying the batteries in a real setup, this test can be carried out without risk using the power HIL approach. In both cases, the PCU must recognize the fault condition and bring the scooter to a standstill and thus to a safe state. The cell temperatures of the virtual battery can also be varied as required without delay. If the cell temperatures are too high, the PCU must limit the output power until the battery returns to a safe temperature.
In reality, it would take hours for the battery cell temperature to warm up or cool down, for example. In simulation, this happens promptly. In addition, this test procedure for different temperatures can even be automated by parameter variation, which reduces the effort and person-hours involved in these tests by over 90%.
Exemplary power HIL test bench. From left to right: device under test cabinet, two cabinets full of high-voltage load modules, rack with real-time system and power supply units
Failure Can Be Simulated at Signal Level Thanks to the Failure Insertion Unit
In addition to faults in the drivetrain, faults can also be mapped at signal level. A Failure Insertion Unit (FIU) can, for example, simulate that the position detection of the throttle twist grip on the handlebars is providing incorrect information to the PCU. The further the throttle twist grip is turned from its rest position, the more target torque should normally calculate by the PCU becomes. The sensor of the throttle twist grip is simulated in the real-time processor,so its rotational angle can be modified arbitrarily. In this case, the PCU must notify the user of the fault and either safely control or stop the vehicle as necessary. Compared to the rotating machine test bench, the power HIL eliminates human error and differences in test results due to test engineer experience.
“All in all, the power HIL offers the possibility of making the component test of the PCU significantly less time-consuming and more flexible,” notes Etsuko Tokunaga. Validation of the PCU at signal level and power level can begin even before all components of the drivetrain have been delivered. This eliminates delays in the development process as far as possible. In the test process itself, the desired operating points can be set both automatically and manually without delay. It is safe to analyze faults because there are no rotating parts and no real batteries are used.
dSPACE provides all components of a power HIL test bench – real-time models, SCALEXIO real-time systems, and load modules – seamlessly from a single source.
Courtesy of Honda
This article has been created in close cooperation with Etsuko Tokunaga, developer in Honda Motor Co., Ltd, Motorcycle Product Development Department, Electrification Development Division.
dSPACE MAGAZINE, PUBLISHED JULY 2025
